Phase II study of metformin for reduction of obesity-associated breast cancer risk: a randomized controlled trial protocol

It is predicted that by 2030 there will be 65 million more obese adults in the USA which will contribute to 492,000–669,000 additional cases of cancer [1]. In addition to cancer, high body mass index (BMI) is a major risk factor for a number of chronic diseases, including type 2 diabetes and cardiovascular diseases [2‐5]. For postmenopausal breast cancer, a high BMI has been reported to increase risk by 30–50 % [6]. The increased risk for postmenopausal breast cancer in women with high adiposity is likely attributed to multiple metabolic disturbances that occur with overweight and obesity including, but not limited to, altered circulating sex steroid hormones, hyperinsulinemic insulin resistance, altered expression and secretion of adipokines from adipose tissue, increased production of pro-inflammatory cytokines, and increased oxidative stress. While intensive weight loss programs can be moderately successful [7], two-thirds of the U.S. adult women are still overweight or obese [8]. Thus, identification of chemoprevention agents that correct the metabolic dysregulation associated with overweight and obesity, even in the absence of weight loss, is a highly attractive strategy for lowering breast cancer risk.

Metformin is a biguanide indicated as an adjunct to diet and exercise to improve glycemic control in adults and children with type 2 diabetes mellitus. It has been used worldwide for over 40 years to treat type 2 diabetes, and was approved in the United States by the Food and Drug Administration as an antidiabetic in 1995 [9]. It is also commonly used off-label for metabolic syndrome [10] as well as polycystic ovary syndrome (PCOS) [11], a condition also characterized by metabolic disturbance. The major mechanism of metformin action in vivo involves suppression of hepatic gluconeogenesis and glucose output, which is associated with a decline in circulating glucose concentration and a secondary decline in insulin levels [12].

Metformin exerts favorable effects on multiple metabolic disturbances that may lead to reduction of breast cancer risk. In diabetics, metformin treatment resulted in favorable changes in circulating levels of leptin and adiponectin [13, 14]. In women with PCOS, metformin has been shown to reduce circulating insulin levels (reviewed by [11]), increase insulin-like growth factor binding protein (IGF-BP) 1 [15, 16], decrease serum testosterone and androstenedione levels [17, 18], and increase serum sex steroid hormone binding globulin [17, 18]. Reduction in serum insulin levels was also observed in non-diabetic breast cancer patients with metformin treatment [19]. Metformin has also been used to treat weight gain induced by antipsychotic medications [20]. It has also been investigated for weight loss in overweight but otherwise healthy individuals with success in some but not all studies, however, these trials had small sample size and design limitations that likely contributed to the inconsistent results across studies (reviewed by [21]).

In addition to indirect effects, metformin may exert a direct effect in mammary tissue through the activation of the AMP-activated protein kinase (AMPK) signaling pathway, leading to an antiproliferative effect and induction of apoptosis [22‐25]. However, metformin is an organic cation at physiological pH. Its cellular uptake generally requires the presence of the cell surface transporters such as organic cation transporters (OCT) 1, 2 or 3. It is not known whether normal human mammary tissue expresses the transporters for metformin to exert its direct effect. In a recent neoadjuvant trial, however, expression of the OCT1 transporter was detected in breast tissue [26].

Some, but not all case control and cohort studies investigating the relationship between diabetes and cancer have found that treatment with metformin appears to substantially reduce the risk for development of cancer in individuals diagnosed with diabetes [27, 28], including lower risk for breast cancer [29‐32]. Some evidence also suggests a role for metformin in prolonging breast cancer survival [33, 34]. Given the retrospective nature of these studies and the possibility that the comparison treatments (such as sulfonylureas or exogenous insulin) may increase risk, randomized, placebo-controlled intervention trials are needed to assess the breast cancer preventive activity of metformin. It is important to note that accumulating evidence suggests that type 2 diabetes and obesity share biological mechanisms for their association with postmenopausal breast cancer which include pathways targeted by metformin (reviewed by [35]). Therefore, metformin could modify breast cancer risk in people with elevated BMI independent of a diagnosis of diabetes.

Recent window-of-opportunity neoadjuvant trials reported clinical activity of metformin in non-diabetic women with operable invasive breast cancer. Two small non-placebo controlled trials showed decreased tumor cell proliferation following a short-term metformin intervention prior to surgery [36, 37]. A considerably larger randomized, placebo-controlled trial (N = 200) did not show an overall decrease in tumor cell proliferation after 4 weeks of metformin intervention [38]. However, subgroup analyses showed a decrease in tumor cell proliferation and serum insulin levels in women with high waist/hip ratio or insulin resistance, suggesting the importance of the metabolic characteristics of the study population [38]. In a recent small trial with early stage breast cancer patients, metformin administration significantly decreased expression of the insulin receptor, phophorylated protein kinase B (PKB), and phosphorylated extracellular signal-regulated kinase 1/2 (ERK 1/2) in tumor tissue [26], suggesting a therapeutic benefit.

An ongoing, large Phase III randomized trial (NCIC CTG MA.32) in over 3,500 early stage breast cancer patients aims to determine whether adding metformin to standard adjuvant therapy will improve invasive disease free survival. Recently published interim data from this trial shows a statistically significant decrease in weight, insulin, glucose, leptin, and CRP at six months in the metformin arm verses placebo [39]. The outcomes of this trial will be important in assessing metformin’s effectiveness on recurrence and survival in breast cancer patients. However, the majority of the study population has received standard adjuvant radiation, endocrine treatment, trastuzumab or other biologics or bisphosphonates prior to or during study treatment, or adjuvant or neoadjuvant chemotherapy prior to study treatment. In general, applicability of the data generated from this trial to a potential role for metformin as a single agent for breast cancer risk reduction in at-risk women with no prior breast cancer history is unclear and requires further research. Our study is one of the first placebo controlled trials to evaluate the effect of metformin on breast cancer risk factors in at risk healthy women.

The overall objective of our study is to evaluate the effect of metformin on biomarkers that have shown strong clinical and scientific relevance to breast cancer risk and have high potential to be modulated by metformin in a cohort of overweight premenopausal women. The primary study aim is to determine the effect of metformin on breast density. Studies suggest that women with dense tissue in more than 60–75 % of the breast are at 4-6-fold greater risk of developing breast cancer than those with no dense tissue [40‐42]. We hypothesize that metformin intervention will reduce breast density because metformin has been shown to decrease breast cell proliferation and modulate the IGF axis [15, 16, 36, 37], which have both been associated with variation in breast density [43‐46]. Breast density will be assessed by magnetic resonance imaging (MRI)-acquired fat-water features, a novel, three-dimensional measure of breast density that will provide more sensitive and quantitative detection of changes in breast density than mammographic measure. The study is enrolling premenopausal women instead of postmenopausal women because premenopausal women have a higher breast density at baseline, which may allow for more notable detection of breast density reduction by metformin. Further, this study population is at increased risk for future postmenopausal breast cancer.

This study is one of the first placebo-controlled, randomized trials to determine metformin effects on multiple metabolic disturbances as well as anthropometric measures in non-diabetic women. In addition, we will explore the application of metabolomics analysis to both plasma and nipple aspirate fluid as a systems biology approach to assess the potential chemopreventive mechanisms of metformin and to understand metabolic features affecting systemic and tissue markers of breast cancer risk. Metabolomic profiling in fluid expressed from the breast may provide information about the breast microenvironment that may not be reflected in plasma. Further, we will also have the opportunity to explore breast-tissue level effects of metformin in a subset of women. Thus, we will have the unique opportunity to explore the associations between metabolic features in nipple aspirate fluid and plasma with markers of breast cancer risk in both plasma and in breast tissue to further our understanding of the underlying biochemical mechanisms affecting these risk markers. Findings from this study will add insight into whether metformin can be used clinically as a pharmacological approach for breast cancer risk reduction in premenopausal women with high adiposity.

An exploratory objective is to determine the effect of metformin intervention on tissue architecture as well as cellular and molecular targets in breast tissue collected in a subgroup of participants. We will also explore whether metformin-induced metabolic changes correlate with changes in markers of breast cancer risk or side effects.

This is a Phase II double-blind, randomized, placebo-controlled trial of metformin to determine its potential effects on recognized and putative markers of breast cancer risk in overweight or obese premenopausal women with elements of metabolic syndrome. Figure 1 illustrates the overall study design.

This is a single institution trial conducted at the University of Arizona Cancer Center (UACC), USA. Women will be recruited through clinic offices, television and radio advertising, printed flyers, social media, and word of mouth.

Randomization is performed using computer-generated random permuted blocks. To retain the blind, metformin and placebo supplies are identified by the randomization number. The study staff will dispense the product to subjects based on the assigned randomization number. None of the staff interacting with subjects will know the link between randomization number and actual product. The code that identifies the product will be kept by the study statistician or the designated data manager.

Unblinding is not expected to occur until all subjects complete the intervention and data entry is complete. If deemed medically necessary, study agents may be unblinded by the Principle Investigator in consultation with the Medical Director in the event of a serious adverse event (SAE).

Subjects will be on agent intervention for 12 months (52 ± 4 weeks). For the first four weeks, 1 metformin (850 mg) or placebo tablet will be taken once daily with food. For the remaining treatment period, metformin or placebo tablets will be taken twice daily with food.

Participants may not use non-study metformin or other biguanides while on study. Medications with potential interaction with metformin (e.g. cationic drugs) will be carefully reviewed and monitored closely while subjects are on study. All medications (prescription and over-the-counter), vitamin and mineral supplements, and/or herbs taken by the participant will be documented on the concomitant medication case report form and include: start and stop date, dose and route of administration, and indication.

A schedule of study procedures is presented in Table 1.

All AEs will be assessed according Common Terminology Criteria for Adverse Events Version 4, followed according to good medical practices, and documented.

The University of Arizona Cancer Center (UACC) Data and Safety Monitoring Board (DSMB) will provide ongoing oversight for this trial. Routine monitoring will be provided by the UACC Quality Assurance/Quality Control (QA/QC) Program to ensure that the investigation is conducted according to protocol design and regulatory requirements. The Principal Investigator will ensure the accuracy, completeness, legibility and timeliness of the data reported in the Case Report Form (CRF). Source documentation supporting the CRF data will indicate the participant’s participation in the trial and should document the dates and details of study procedures, AEs, and participant status.

Our sample size ensures adequate power to test all primary and secondary outcomes. We assume an initial sample size of 75 women per group and allow the possibility of up to 20 % dropout for the 12-month follow-up measurement, resulting in at least 60 evaluable women per group. To determine the detectable effect size, we use a simplified comparison of change between 12 months versus baseline, assuming a two-sided α of 0.05 and 80 % statistical power. For the fat-water MRI-derived change in breast density, we will be able to detect a decrease of 0.516 standard deviation (SD) units in the metformin treated women, equivalent to a 10 % decrease in breast density as determined by standard mammography. For the secondary outcomes (such as systemic hormones, cytokines, and weight), we will also be able to detect a difference of 0.516 SD units. For metabolomic analysis that compares thousands of metabolites, we approximated the effect size by setting a more stringent alpha level of 0.005. For plasma metabolomic analysis we will be able to detect an effect size of 0.667 SD units, while for NAF metabolomic analysis we will be able to detect an effect size of 0.848 SD units (assuming at least 65 % of the women will provide sufficient NAF).

The primary study endpoint is to compare the change in breast density as measured by fat-water MRI (when measured at baseline, 6 and 12 months) between metformin and placebo groups. Additional endpoints such as systemic hormone and cytokines, body weight, waist circumference, and waist-hip ratio will be measured at baseline, 6 and 12 months. Analysis for all study endpoints will be based on a linear mixed-effects model for the observed values across time, to adjust for the correlation among measurements within the same woman. The main effects in the model will be time (0, 6, 12), treatment group (metformin versus placebo), and the interaction between time and treatment group. The time parameter tests if there is a change among placebo treated women, while the group-by-time interaction tests if the change in metformin treated women differs from that in the placebo group. We expect that a simpler covariance structure, such as compound symmetry, will be adequate for repeated measurements within the same woman (since the measurements are equally spaced). Alternatively, we will compare correlation structures for the longitudinal measurements using Akaike’s information criterion. All endpoints will be assessed for normality, and transformations (such as a logarithmic transformation) will be used as necessary to reduce skewness prior to statistical analysis.

For the metabolomic aim, pre- to post-intervention changes in all detectable compounds will be determined and compared between treatment groups using two-sample t-tests. An estimate of the false discovery rate (q-value) will be calculated [51] to adjust for the multiple comparisons that normally occur in metabolomic-based studies. A low q-value (q ≤ 0.10) is an indication of high confidence in a result. While a higher q-value indicates diminished confidence, it does not necessarily rule out the significance of a result. Other lines of evidence may be taken into consideration when determining whether a result merits further scrutiny. Such evidence may include a) significance in another dimension of the study, b) inclusion in a common pathway with a highly significant compound, or c) residing in a similar functional biochemical family with other significant compounds.

No interim statistical analyses are planned for this Phase II trial. Accrual, data collection, and any AEs will be monitored on a regular basis. The final dataset will be available to all study investigators.